algorithm
Latest
An algorithm can tell if your face is forgettable
Some faces are more memorable than others. The brain processes visual cues to decide if a face or an image will stay lodged in the memory bank. What if a network could be trained to imitate that response? You could potentially alter the visuals for a greater impact. A team of researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) has built MemNet, a deep learning based algorithm that predicts the "memorability" of your photographs almost as well as the human brain.
The After Math: With great power
Well, this week as been rather terrible. With all the death and mayhem both at home and abroad, it's enough to make anyone feel rather helpless. It's times like these that we have to force ourselves to remember there is still a great deal of good left in the world. From life-saving medical advancements and clean energy promises to superior image scanners and kick-assier video games, this week's selection shows that humanity isn't all bad (just mostly).
Computers learn how to spot hidden facial expressions
Machines are good at spotting obvious emotions like smiles, but they're not so hot at detecting the extremely brief microexpressions that reveal when people are covering up their true feelings. They may have a keener eye in the future, though: researchers have developed a computer vision algorithm that magnifies facial expressions, making it possible to catch the tiniest bit of displeasure or surprise. While some humans have a knack for spotting these subtle cues, the algorithm is far more effective in early tests -- you likely wouldn't fool the computer into thinking everything was hunky dory.
Tinder hopes its new features will improve your odds
'Super likes' be damned, Tinder is fleshing out its addictively swipey profile cards with more information aimed at connecting you with the person of your dreams / evening. As well as "intelligently" adding relevant information (including work history and education) on each suitor's audition card, the matching-making app says it's improved its learning algorithm to "drive more compatible matches." Machine learning will assess what Tinder users have been doing with the app to create an algorithm to help improve the chances of love. How exactly that'll help your odds, when everyone will still be swiping in one direction or another, remains to be seen -- we've asked Tinder for clarification.
Mitsubishi's using AI to save distracted drivers from themselves
There's no doubt that self-driving vehicles will play a huge part in our automotive future, but until they do, ensuring that human drivers stay safe on the road remains the top priority for car companies. Japanese manufacturer Mitsubishi Electric, one of the Mitsubishi Group's many subsidiaries, reckons more can be done to keep a driver's focus on the road, so it's developed a new technology that can detect when someone is distracted or feeling tired.
Watch code and projections bring a paper sculpture to life
Paper art doesn't have to be flat and lifeless... just ask Aristides Garcia. The artist recently created an interactive sculpture, Tesela, that uses a combination of 3D projection mapping and tesselation algorithms to cast real-time, viewer-influenced patterns over 103 paper pyramids. The effect is a bit hypnotic, as you'll see below -- it's as if the paper has suddenly become a living landscape. You sadly can't see this in person at the moment (Garcia debuted it at a Berlin exhibit in August), but it still shows that the right technology can liven up just about anything, even if it's made from dead trees.
Animated code art uses all of its colors just once
You probably know that the screen on your computer or phone can display millions of colors, if not more. However, have you wondered what it would look like if you tried to represent all of those colors in a single piece of art? Well, you're looking at it. Qubit researcher and math guru Mike Swarbrick Jones has posted a code-driven animation that shows all the colors in a 24-bit RGB palette exactly once. The technique (which relies on mapping colors to voxels, or 3D pixels) produces a kind of "rainbow smoke" that, as you can see in the clip below, is rather hypnotic -- it's tempting to watch it on a loop and meditate. While this won't produce a masterpiece, it's proof that a good idea and the right calculations can lead to some truly eye-catching (not to mention mind-bending) visuals.
Your phone knows if you're bored by how much you use it
Researchers at Barcelona's Telefonica Research lab have developed a smartphone-based algorithm that determines a user's level of boredom based on how much they're using the device. The algorithm also takes a number of factors such as time of day and how long it's been since receiving a call or text into account as well. With it, the researchers were able to accurately gauge a user's level of boredom 83 percent of the time.
Smoother movements help robots save a lot of energy
Eliminating the herky-jerky movements of robots isn't just good for comforting nervous humans... it helps the robots, too. Researchers have developed smooth movement algorithms that slow the acceleration and deceleration of robots, saving as much as 40 percent of the energy they'd normally use. The trick is to order tasks in a way that lets robots move at their own pace without colliding into each other. Factory robots typically rush through tasks in a rigid order, only to wait for their fellow automatons to catch up. Here, they're more flexible as to when and how quickly they get things done.
Computers can categorize buildings into architectural styles
Even if you've never heard of "Byzantine," you can probably tell a Byzantine church from a Gothic one. Judging style differences is nearly impossible for a computer, however, and researchers from the University of Massachusetts want to fill in that gap. They used geometric matching, crowdsourcing and machine learning to teach an algorithm how to spot similar styles in buildings, furniture and other objects. That's something that could be incredibly useful for historians with mountains of photo archives, or game designers who need to auto-fill a level with historically accurate furniture.
Facebook algorithm can recognize people if it can't see their face
A number of companies have developed photo software for facial recognition, but what happens when your face is partially hidden? What if it's completely covered up? Facebook's artificial intelligence lab developed an algorithm that remedies the issue by picking out folks with other clues. Instead of using facial features, the software can identify people using things like hair style, pose, clothing and body type. Of course, a tool like this could lend a hand in a photo app like Facebook Moments or even Google's revamped Photos software. However, it also raises privacy questions when you can be identified in a snapshot even if your face is concealed, especially if you're trying to remain hidden on purpose. Facebook's algorithm is pretty good too, identifying people with an 83 percent success rate in tests, so we'll be curious to see if it makes its way into the social network's photo galleries in the future. [Image credit: David Paul Morris/Bloomberg via Getty Images]
Facebook and Google get neural networks to create art
For Facebook and Google, it's not enough for computers to recognize images... they should create images, too. Both tech firms have just shown off neural networks that automatically generate pictures based on their understanding of what objects look like. Facebook's approach uses two of these networks to produce tiny thumbnail images. The technique is much like what you'd experience if you learned painting from a harsh (if not especially daring) critic. The first algorithm creates pictures based on a random vector, while the second checks them for realistic objects and rejects the fake-looking shots; over time, you're left with the most convincing results. The current output is good enough that 40 percent of pictures fooled human viewers, and there's a chance that they'll become more realistic with further refinements.
Facebook tweaks your newsfeed by how long you read each post
In an unsurprising revelation, it turns out your Facebook news feed is watching you almost as much as you watch it. The Menlo Park-based company announced today that it is "improving" its news feature by taking into account not just whether someone liked or commented on an article but also by how much time they spent reading it. "Just because someone didn't like, comment or share a story in their News Feed doesn't mean it wasn't meaningful to them," Facebook explains."There are times when, for example, people want to see information about a serious current event, but don't necessarily want to like or comment on it."
Computer algorithm picks the world's most creative art
Who would you trust to determine history's most creative art? A room full of seasoned critics? Rutgers University researchers think a machine can do the job. They've developed a computer vision algorithm that ranks the creativity of art based on how similar it is to earlier works in terms of everything from color and texture to the presence of familiar objects. The code treats art history as a network -- groundbreaking pieces are connected to later derivatives, and seemingly unique content may have a link to something produced in the distant past.
Wolfram's new website can identify objects in your photos
Wolfram Research can already do some pretty cool things, like answer Twitter questions and spot overhead flights. Now, the maker of the Mathematica programming language and Alpha knowledge engine can perform another trick: figuring out what's in a photo. The Wolfram Language Image Identification Project can make out about 10,000 common things, including animal species, gadgets and household objects. It uses a database of around ten million images to perform the trick, which Stephen Wolfram figures "is comparable to the number of distinct views of objects that humans get in their first couple of years of life."
MIT can fix pictures taken through your window... kinda
Take a picture through a window and you'll often find you've captured more of your own reflection than the scene outside. You can solve the problem with a black cloth and a polarizing filter, but that's not ideal for the majority of smartphone snappers out there. That's why researchers at MIT are about to present a new software-based solution that, they believe, can "fix" the problem, but only if the window that you're shooting through is double-glazed.
MIT gave exploring robots a way to plan underwater missions
Forget those teensy deep-sea submersibles cradling crews of brave scientists -- the future of underwater exploration might be led by robots that can do their own thing. MIT engineers, led by professor Brian Williams, cooked up a system that lets autonomous underwater drones figure out and act on the nitty-gritty details of their missions without the need for meticulously laid-out plans.
Here's how 'Minecraft' creates its gigantic worlds
Have you wondered how Minecraft can produce massive worlds that are still chock-full of little details, like elaborate cliff faces and waterfalls? PBS' Game/Show is more than happy to explain in a new video. As you'll see below, Mojang's game relies on procedural generation, which automatically creates environments and objects that are at once random, but guided by rules that maintain a consistent logic. Mountains are always rocky and sprinkled with snow, for example, while the low lands are typically full of grass and trees.
Algorithm determines which rappers have the slickest rhymes
The days of arguing over the worth of your favorite rappers might soon come to an end. Data mining student Eric Malmi has built Raplyzer, an algorithmic program that gauges the average length of a rap or hip-hop star's multi-syllable rhymes (the key to the "dopest flows," Flocabulary says) and ranks that person accordingly. Based on this math, the champions are a mix of veterans and relative newcomers. Wu-Tang Clan's Inspectah Deck is on top, while big names like Rakim, Earl Sweatshirt and ASAP Rocky are near the front.
MIT's light-up robot garden teaches you how to code
If you're teaching kids how to code, what do you do to show that software makes an impact in the real world? MIT has a clever idea: a robot garden. The project lets you control a grid of Arduino-linked "plants" through programming that makes them blossom and light up in pretty (and occasionally mesmerizing) ways. It'll even teach the virtues of distributed computing -- you can tell these leafy robots to bloom or change color in algorithm-driven sequences. The garden is just a demo for now, but it'll eventually turn into an easy-to-replicate curriculum for students who'd otherwise have to settle for seeing their results on-screen. [Image credit: Jason Dorfman, CSAIL]
